Vous êtes sur la page 1sur 8

99

Early pregnancy human chorionic gonadotropin (hCG) isoforms


measured by an immunometric assay for choriocarcinoma-like hCG
G Kovalevskaya1, S Birken2, T Kakuma3 and J F OConnor1,4
1

Irving Center for Clinical Research, Columbia University College of Physicians and Surgeons, 630 W 168th St., New York 10032, New York, USA

Department of Medicine, Columbia University College of Physicians and Surgeons, 630 W 168th St., New York 10032, New York, USA

New York Hospital-Cornell Medical Center, White Plains, New York 10605, New York, USA

Department of Pathology, Columbia University College of Physicians and Surgeons, 630 W 168th St., New York 10032, New York, USA

(Requests for offprints should be addressed to J F OConnor, Department of Pathology and Irving Center for Clinical Research, Columbia University College of
Physicians and Surgeons, 630 W 168th St., PH10305, New York 10032, New York, USA)

Abstract
Human chorionic gonadotropin (hCG) exhibits molecular
heterogeneity in both its protein and carbohydrate
moieties. This communication describes changes in hCG
isoforms detected directly in clinical samples. These isoforms, quantified in blood or urine specimens, show a
progression of change throughout normal pregnancy. Early
pregnancy produces a type of hCG that resembles, in
terms of immunoreactivity, a major form of hCG excreted
in choriocarcinoma. The isoforms predominate for the first
56 weeks of gestation and then diminish, being replaced
with the hCG isoforms which predominate throughout
the remainder of pregnancy. The alteration in hCG
isoform content occurs in both blood and urine. The
progression of isoforms is best delineated by calculating the

Introduction
Human chorionic gonadotropin (hCG) is produced by the
placental trophoblast of normal pregnancy, and in gestational trophoblastic disease (hydatidiform mole and choriocarcinoma) (Hussa 1987). It is also produced in much smaller
quantities by the pituitary (Birken et al. 1996) in both preand postmenopausal women and in men (Odell et al. 1990)
and also in many non-gestational malignant tumors (Hussa
1987).
Structural variations in both the protein and carbohydrate portions of the hCG molecule have been well
established. They may occur as a concomitant of its origin
(Wide & Hobson 1987, Birken et al. 1996), or metabolic
transformation (Nisula et al. 1989). The most prominent
differences documented to date are between the hCG
produced by choriocarcinoma and that of mid-first trimester normal pregnancy, with choriocarcinoma-derived
hCG having a generally higher content of more highly
branched oligosaccharides (Cole 1987, Elliott et al. 1997).
Even in normal pregnancy, multiple isoforms of hCG
exist and their relative amounts vary as pregnancy

change in the ratio of the two forms, as many hCG assays


either do not detect or fail to discriminate among these
isoforms. An analogous pattern of hCG isoforms was
observed in patients with in vitro fertilization pregnancies.
hCG isolated from the pituitary displayed binding characteristics similar to those of the hCG derived from normal
pregnancy urine. The early pregnancy hCG isoforms
appear to have a differential expression in normal pregnancy as opposed to pregnancies which will not carry to
term, suggesting that a determination of the relative
balance of hCG isoforms may have diagnostic application
in predicting pregnancy outcome.
Journal of Endocrinology (1999) 161, 99106

progresses (Wide & Hobson 1987, Skarulis et al. 1992,


Diaz-Cueto et al. 1994, 1996, Wide et al. 1994). Many of
these determinations are based on isoelectric focusing
(IEF) separation of the isoforms. This technique separates
molecules based on their net charge, which in the case of
glycoprotein hormones only reflects the degree of terminal
carbohydrate sialylation (or less commonly, sulfation), and
not major structural differences (Ulloa-Aguirre et al. 1990,
Berger et al. 1993, Diaz-Cueto et al. 1994). However,
these studies of microheterogeneity (by microheterogeneity, we refer to the permutations displayed by the
carbohydrate groups) have not proven to have clinical
utility, since they are both too expensive and labor
intensive to perform routinely. Other studies have demonstrated, by analysis of carbohydrate structure and sugar
content, differences in the carbohydrate moiety of normal
pregnancy hCG as it progresses to term (Skarulis et al.
1992, Nemansky et al. 1998).
This report describes the observation of the progressive
changes of hCG isoforms throughout pregnancy by direct
assay of clinical samples of urine or serum, using a
combination of two IRMAs. One of these, a new IRMA,

Journal of Endocrinology (1999) 161, 99106


00220795/99/01610099  1999 Society for Endocrinology Printed in Great Britain

Online version via http://www.endocrinology.org

100

G KOVALEVSKAYA

and others

Early pregnancy hCG

is unique in that it is sensitive to changes in carbohydrate


structure and can measure very early isoforms of hCG
(OConnor et al. 1998), whereas the second assay, a
commonly used format, detects isoforms of hCG which
predominate later in pregnancy. This observation takes on
an added significance because of reports that certain
pregnancy disorders (e.g. Down syndrome, pre-eclampsia,
early pregnancy loss) are characterized by abnormal hCG
levels (Bogart et al. 1987, Heinonen et al. 1996, Hurley
et al. 1996, OLeary et al. 1996, Onderoglu & Kabukcu
1997, Haddow et al. 1998). We have evidence that
pregnancies which will not carry to term have an easily
recognized difference in production of hCG isoforms
(OConnor et al. 1998). Thus, hCG isoform production
and distribution in pregnancy may prove to have a
diagnostic role in the evaluation of pregnancy status. The
method of direct IRMA analysis of early pregnancy
isoforms described in this report is suitable for large scale
studies using either serum or urine.

Materials and Methods


Hormones
The non-nicked hCG isolated from the CR127 preparation of hCG was used as a standard in both assays
(Birken et al. 1993). The pituitary hCG was isolated as
described (Birken et al. 1996). C5, a 100% nicked hCG
having extra sugars on both N- and O-linked carbohydrate moieties, purified from the urine of a choriocarcinoma patient (Elliott et al. 1997), was supplied by Dr
L Cole (Yale University School of Medicine). Although
the C5 immunogen used in the development of B152
antibody was 100% nicked hCG isoform (i.e. had cleavages in the peptide backbone of loop 2 of the subunit)
the antibody did not discriminate nicked from non-nicked
hCG (OConnor et al. 1998).
The same serial dilutions of non-nicked hCG, pituitary
hCG and C5 were used for binding characterization in
hCG assays. Hormone concentrations of initial stock
standard solutions were determined by amino acid analysis.

B207* assay) in coating buffer (02 M bicarbonate, pH 95)


overnight at 4 C. The coating antibody solution was
aspirated, the plates washed (wash solution: 09% NaCl,
005% Tween 20) and blocked with a 1% solution of BSA
in PBS with 01% sodium azide. Following incubation
with the BSA solution (minimum 3 h at room temperature), the blocking solution was removed, the wells again
washed with wash solution and 200 l/well of the appropriate hCG standards were added in phosphate buffer B
(PBS with 01% bovine gamma globulin and 01% sodium
azide). After overnight incubation at 4 C, the plates were
again aspirated and washed. The 200 l of 125I-labeled
antibody (50 000100 000 c.p.m.) were added to the
wells, which were again incubated for 24 h at 4 C. The
tracer was aspirated, the plates washed with wash solution,
the individual wells placed in glass tubes and the radioactivity determined in a Packard Cobra gamma counter
(Downers Grove, IL, USA). Doses were determined by
interpolation from a smoothed spline transformation of the
data points.
All samples were stored frozen at20 C prior to assay.
Because extreme values of sample pH may interfere with
antibody binding, the urine pH was adjusted with 10 M
Tris (pH 90), 50 l/ml urine, prior to assay, so that the
final pH was in the range 7074 (OConnor et al. 1988).
Intra-assay variation was 6% for both assays, inter-assay
variation was 12% for B109B108* and 13% for B152
B207* assays. Sensitivity (least detectable dose) defined as
+2.. from the zero calibrator, was 10 fmol/ml for the
B109B108* assay and 22 fmol/ml for the B152B207*
assay.
The only mapping of the epitope sites thus far has
established that the B108 and B207 tracer binding site
corresponds to site II, the same site recognized by the SB6
antiserum (Vaitukaitis et al. 1972). The B152 capture site
involves the carboxy-terminal portion of the hCG
subunit, since B152 will not bind simultaneously with a
carboxyterminal peptide-specific monoclonal antibody. In
the B109B108* assay, B109 requires that the and
subunits both be present in the hCG molecule, whereas
B152 will bind either the intact, heterodimeric hCG
molecule or its free subunit (OConnor et al. 1994,
Birken et al., unpublished observations).

IRMAs
The methodology used in the construction and validation
of the B109B108* assay has been fully described elsewhere (OConnor et al. 1988); the first antibody is the
capture antibody, the second, with an asterisk, is a
radioiodinated detection antibody. The B152B207* assay
has also been characterized (OConnor et al. 1998). Both
assays were performed with a slight modification of the
published procedure: the capture antibody was adsorbed
onto the wells of microtiter plates (Immulon IV,
Dynatech, Chantilly, VA, USA) by incubating a 5 g/ml
solution (B109B108* assay) or 25 g/ml solution (B152
Journal of Endocrinology (1999) 161, 99106

Patient samples
Urine samples from in vitro fertilization (IVF) patients were
a gift from Dr L Cole. They included spontaneous abortion
(n=14, range of gestational age 1841 weeks from
embryo transfer (ET)), ectopic pregnancies (n=7, gestational age 2340 weeks) and normal pregnancy controls
(n=65, encompassing the range 0654 weeks from ET).
Some of the normal pregnancy urine samples throughout
the pregnancies were also obtained from Dr Cole. Others
were obtained from the clinical practice of collaborating
physicians at Columbia Presbyterian Medical Center

Early pregnancy hCG

G KOVALEVSKAYA

and others

(CPMC) (total n=103). Matched serum/urine samples


from the first (n=12) and the third (n=11) trimesters were
provided by Dr A Kelly at CPMC. Trophoblast disease
serum (n=17) and urine (n=28) samples were obtained
from Dr Cole, but had been collected by Dr E Newlands
(Charing Cross Hospital, London, UK). All specimen
collection protocols were approved by the appropriate
Institutional Review Board.
Statistical analysis
Polynomial regression models of log-transformed hormone
ratios were used to describe the relationship between the
change in ratio as a function of gestational age in normal
pregnancy. A paired t-test was used to evaluate the
relationship between matched serum and urine hormone
ratios. Analysis of covariance (ANCOVA) was used to
describe the time-adjusted relationship of hormone values
in ectopic pregnancy and spontaneous abortion to those of
normal pregnancy.
Results
Immunoreactivity of different forms of hCG in the two
IRMAs
The relative binding of three different forms of hCG
(urinary hCG, pituitary hCG and choriocarcinoma hCG
C5) has been characterized in the two hCG assays (Fig. 1).
Urinary non-nicked hCG and pituitary hCG are recognized nearly equally well by the two IRMAs, while C5
recognition is quite different. The B152B207* assay is
more sensitive to C5, which is to be expected because
B152 antibody was developed and selected on the basis of
higher affinity to C5. Urinary non-nicked hCG is purified
from the CR127 preparation of pooled normal pregnancy
hCG. Conversely, C5 is recognized with lower affinity by
the B109B108* assay, which has primary specificity for
the hCG isoforms of later pregnancy.
The B152/B109 ratio measured in urine samples throughout
the pregnancy
The relative concentrations of hCG isoforms in 103
normal pregnancy urine samples (539 weeks post last
menstrual period (LMP)) were determined by the two
IRMAs (B152B207* and B109B108*). Both because
of the wide range of hCG concentrations in different
samples, even at the same gestational age, and because
neither of the assays is totally specific for the two (or more)
families of hCG isoforms present, we find that presenting
the data as a ratio of the observed two isoform groups more
clearly delineates the change in isoform content as pregnancy progresses. This calculated ratio is shown in Fig. 2.
In weeks 58 of pregnancy, the ratio of B152/B109

Figure 1 Binding curves for three hCG types in the B152B207*


assay (upper panel) and the B109B108* assay (lower panel).

isoforms ranged from 62 to 13, indicating a predominance of the B152 isoform(s) in early pregnancy. During
the 1012 week period, the ratio ranged from 10 to 02,
indicating that an inversion in hCG isoform content is
occurring as pregnancy progresses. This decline in the
ratio continues, ranging from 054 to 008 in the 1518
week period and reaching an inflection point at 29 weeks.
Journal of Endocrinology (1999) 161, 99106

101

102

G KOVALEVSKAYA

and others

Early pregnancy hCG

Figure 3 Box plot of the B152/B109 ratio for pregnancy matched


serum/urine at 56 weeks of gestational age (n=12), at 3639
weeks of gestational age (n=11) and in JAR cell supernatant.
Boxes extend to the 25th and 75th percentile. The upper and
lower symbols indicate the 90th and 10th percentile respectively.
A solid line inside the box marks the value of the 50th percentile.
Table 1 Analysis of the B152/B109 ratio in serum and in urine in
the first vs third trimesters of pregnancy

Figure 2 Ratio of hCG isoforms measured by the B152B207*


and B109B108* assays in normal pregnancy urine (n=103) at
different gestational ages. (Regression curve and 95% confidence
intervals are shown, r2=079.) An inflection point in the curve
occurs at approximately 29 weeks.

At that time, the ratio reached a value of around 006, after


which the ratio displayed a rise to a range from 020 to
007 in the 37395 weeks of gestation time period.
Statistical analysis involved fitting the log-transformed
ratio data to second and third order polynomial regression
models. Since the third order term was not significant
(likelihood ratio 2(1)=132, P=025), the second order
model was used (r2=0793). The log B152/B109 ratio
reached an inflection point at LMP=29 weeks, based on
this model.
The B152B207* values reflect a measurement of the
B152 isoform in terms of later pregnancy hCG equivalents, not in absolute quantities. It must be emphasized
that the absolute concentrations measured in the B152
assay cannot be compared with the results of the B109
assay on an equimolar basis, since the potency of the
hyperglycosylated isoform is much higher in the B152
assay vis-a`-vis the standard, i.e. normal later first trimester
pregnancy hCG. The actual molar values of this isoform
are of the order of tenfold less than those recorded in the
assay. For this reason, we have chosen not to analyze
absolute molar quantities of the two analytes, but only the
ratio of the two measurements.
The B152/B109 ratio in matched serum/urine samples in the
first and third trimesters of pregnancy compared with hCG
from JAR cells
The B152/B109 ratio in serum is analogous to that found
in matched urine samples and undergoes a similar change
Journal of Endocrinology (1999) 161, 99106

Measure
Serum, ratio B152/B109
Serum, log(ratio B152/B109)
Urine, ratio B152/B109
Urine, log(ratio B152/B109)

t-test=(df)

t(11) =665
t(23) =2161
t(11) =464
t(157)=1685

00001
00000
00007
00001

as pregnancy progresses (Fig. 3). The B152/B109 ratio in


the cell supernatant from JAR cells (a choriocarcinomaderived cell line) was similar to that of early pregnancy.
The B152/B109 ratios of both serum and urine hCG
concentrations are significantly higher in the first trimester
as compared with the third trimester of normal pregnancies (Table 1). Significant differences between serum
and urine hCG concentration ratios as well as logtransformed ratios in early (56 weeks) and late (3639
weeks) gestation were evaluated by paired t-tests (Table
2). In both the first and third trimesters, urinary B152/
B109 ratios were significantly higher than serum ratios,
indicating that there was a preferential clearance of the
B152-recognized isoform into urine, regardless of the
relative concentrations of the two isoforms.
The B152/B109 ratio in urine samples from IVF patients
In urine samples from IVF patients (14 weeks post ET)
the B152/B109 ratio was again between 2 and 8 and
decreased as pregnancy progressed (Fig. 4), similar to that
observed in natural conceptions. The effect of pregnancy
duration with respect to outcome variables could best be
represented by a linear or quadratic function. ANCOVA
models including the second order week were fitted to the
general equation: Outcome=(effect of time post
ET)+(effect of diagnosis). After an appropriate ANCOVA
model was determined, the least square means (adjusted
for week post ET effect) were compared among the

Early pregnancy hCG

G KOVALEVSKAYA

and others

Table 2 Analysis of the B152/B109 ratio in serum vs urine in the first and third trimesters
of pregnancy

Gestational age (weeks)


56
3639

Measure

Paired t (df)

Ratio B152/B109
Log(ratio B152/B109)

t(11)=325
t(11)=625

00077
00001

Ratio B152/B109
Log(ratio B152/B109)

t(10)=547
t(10)=714

00003
00001

concentration in serum was 202 fmol/ml in the B152


B207* assay, with a corresponding value of 148 fmol/ml
in the B109B108* determination. Six of seventeen
samples in serum had detectable levels, with four out of six
having a higher value in the B152B207* assay. Of the
15 out of 28 positive urine samples, however, 14 out
of 15 had higher levels in the B152B207* assay than
in the B109B108* assay, with the highest hCG value
being 20 000 fmol/ml in the B152B207* assay and
18 715 fmol/ml in the corresponding B109B108* assay.
Due to the small sample size, no statistical treatment was
performed on these data, but even in these post-treatment
patients the B152/B109 ratio was >1, which corresponds
to the early pregnancy hCG isoform ratio.
Discussion

Figure 4 Ratio of hCG isoforms measured by the B152B207*


and B109B108* assays in the urine of IVF patients (n=65).
(Regression curve and 95% confidence intervals are shown,
r2=059.)

normal pregnancy, ectopic pregnancy and spontaneous


abortion populations (Table 3). The log-transformed values of both B109B108*- and B152B207*-measured
hCG forms discriminated both ectopic pregnancy and
spontaneous abortions from normal pregnancy (P=
00001). The ratio of the log-transformed values discriminated abortion from normal pregnancy (P=0016). However, neither the ratio of B152/B109 nor the log of that
ratio discriminated either of the pregnancy disorders from
normal pregnancy.
hCG analysis of trophoblastic disease samples
Trophoblast disease serum (17 samples) and urine
(28 samples) were obtained from patients post therapy and
hence contained low hCG levels. Due to limited amounts
of sample, all of these specimens were run at a 1:10 initial
dilution. hCG levels in serum were low. The highest hCG

We have developed a method to directly profile changes


of hCG isoforms in serum or urine throughout pregnancy.
Two IRMAs for hCG are employed, each based on
monoclonal antibodies to different hCG epitopes. The
B109B108* assay is a commonly used intact hCG assay
to the heterodimeric-dependent epitope. A new assay,
B152B207*, is most likely sensitive to the carbohydrate
portion of hCG carboxy-terminal peptide. The same
standard non-nicked hCG was used in both assays. Nonnicked hCG was employed, since the B109 assay reacts
poorly with nicked forms of hCG, while the B152 assay
does not discriminate between nicked and non-nicked
forms of the hormone. The B152 assay detected with
greatly enhanced sensitivity hCG isoforms which appear
earlier in pregnancy than isoforms measured by the B109
assay (OConnor et al. 1998). Prior to development of the
new IRMA system described in this report, it was not
possible to readily discern the changes in hCG isoforms
from very early pregnancy to mid pregnancy. The only
available procedure for examining these changes was IEF
of every patient specimen followed by immunoassay of
every focused fraction (Ulloa-Aguirre et al. 1990, Berger
et al. 1993). The IEF pattern reflects the heterogeneity of
the charged sugar, sialic acid, which varies with the
multi-antennary structures of the carbohydrate moieties in
which sialic acid is the terminal sugar. Although we do not
yet know the precise nature of the isoform epitopes being
Journal of Endocrinology (1999) 161, 99106

103

104

G KOVALEVSKAYA

and others

Early pregnancy hCG

Table 3 IVF patients: analysis of covariance of hCG isoforms among normal pregnancy,
ectopic pregnancy and spontaneous abortion as a function of gestational age
Adjusted r2c

Pairwise differenced

Diagnosis
Fe

Outcome
Log(ratio B152/B109)a

051

089

056

2133

00001

Log(B152)/log(B109)b

045

434

0016

Log(B152B207*)b

050

2694

00001

Log(B109B108*)

041

None
Normal pregnancy vs
abortion and ectopic
Normal pregnancy vs
abortion
Normal pregnancy vs
abortion and ectopic

ANCOVA model with 2nd order polynomial coefficient (or parameter).


ANCOVA model with only 1st order (linear) coefficient.
Adjusted r2 is an r2 adjusting number of coefficients on the ANCOVA model so that comparisons of
two r2 with different ANOVA models with different numbers of coefficients are meaningful.
d
Pairwise difference is based on t-test comparing the least-square means of outcome variables (after
adjusting effect of week of ET).
e
Degrees of freedom (dfl, df2) for F-test are 2,82) for a model with only linear coefficient and (2,81) for
a model with both linear and 2nd order coefficient.
b
c

measured, the evidence for carbohydrate discrimination is


based upon the hyperglycosylated structure of the immunogen, C5, used to develop the B152 monoclonal
antibody and the antibodys reactivity with the hCG
isoforms found in the JAR choriocarcinoma cell line.
C5 hCG was isolated from a choriocarcinoma patient and
has been thoroughly characterized as to its protein and
carbohydrate content and structure (Elliott et al. 1997). It
has been shown that C5, and hCG from other choriocarcinoma subjects, differ in the protein moiety mainly by the
presence of an increased number of nicked sites and by
increased glycosylation relative to the hCG of normal
pregnancy. In comparison with the hCG of normal
pregnancy, choriocarcinoma-derived hCG has increased
fucosylation of the N-linked biantennary oligosaccharides
in the subunit. In addition, the O-linked oligosaccharides in preparation C5 (a form of hCG produced from
a single patient with choriocarcinoma) has a 100% tetrasaccharide core on the carboxy-terminal region of the
subunit. Normal mid pregnancy hCG has only 1020% of
this structure (Elliott et al. 1997). These observations, plus
our own determination that the hCG synthesized by the
JAR choriocarcinoma cell line provides a B152/B109
isoform ratio similar to that observed in early pregnancy,
lead us to the conclusion that, in very early pregnancy, the
developing trophoblast secretes an isoform of hCG which
resembles that produced in choriocarcinoma.
We have also tested recognition of pituitary hCG, since
its N-Asn carbohydrates differ somewhat from those of
placental hCG, bearing a closer resemblance to those of
human luteinizing hormone, which have both sialic acid
and sulfate groups (Birken et al. 1996). The carbohydrate
structure of the carboxy-terminal portion of pituitary
hCG is not yet known. Since B152 did not recognize any
Journal of Endocrinology (1999) 161, 99106

substantial differences between pituitary and placental


hCG (Fig. 1), differences in N-Asn recognition are
unlikely. In terms of the carboxy-terminal carbohydrates,
it appears that pituitary and placental hCG (mid pregnancy
isoforms) may be similar, assuming the O-linked carbohydrate on the C5 antigen is part of the epitope of B152.
Even in normal pregnancy, the hCG values obtained
vary widely according to the characteristics of the immunological reagents employed (Cole & Kardana 1992,
Cole et al. 1993). We hypothesize that the two assays
described in this report primarily detect hCG isoforms at
opposing ends of this spectrum, each primarily recognizing
a subset of closely related molecules in the continuum of
early to later pregnancy hCG molecular forms.
We have retained the use of normal pregnancy hCG as
the standard in the B152B207* assay, despite its decreased affinity in this antibody configuration. The reasons
for this include the limited and unrenewable supply of C5
(which was isolated from the urine of a single patient) and
the variability in data which would result from investigations using different standards. The consequences of this
choice are that the early pregnancy hCG isoforms have
markedly increased immunopotency over that of normal
pregnancy and hence their molar quantities are overestimated in this assay. We use this difference in affinity to our
advantage by employing a ratio of the molar results of two
assays (B152 and B109). Either assay taken alone obscures
this change due to the wide excursion of hCG values
which occur in normal pregnancy.
Others have documented progressive changes in hCG
isoforms throughout pregnancy. Skarulis et al. (1992)
found that the fucose content of both intact hCG and also
its free subunit increased as pregnancy progressed.
Diaz-Cueto et al. (1996), investigating the IEF pattern of

Early pregnancy hCG

circulating hCG throughout pregnancy, found that in


early pregnancy more than 80% of the hCG isoforms were
acidic. This fraction decreased to less than half (47%) late
in the third trimester (Diaz-Cueto et al. 1996). In contrast,
Wide & Hobson (1987) found that the hCG of early
pregnancy was more choriocarcinoma-like by virtue of its
greater biological activity than the hCG of normal pregnancy. Fein et al. (1980), in a study which employed gel
filtration, determined that first trimester hCG was a larger
size than that of the third trimester. Treatment with
exoglycosidases eliminated the size differential, indicating
that the first trimester hCG was more highly glycosylated
(Fein et al. 1980).
A significant number of spontaneous abortions and
ectopic pregnancies occur in IVF pregnancies. We did not
find a difference in the ratio of the isoforms between either
of these two categories as compared with normal controls,
possibly a consequence of low statistical power. However,
a significant difference was found between the B152 hCG
isoforms levels in normal pregnancy and spontaneous
abortion. This supports our previous finding in early
pregnancy loss, where diminished or absent levels of the
B152 isoforms characterized an early pregnancy loss
(OConnor et al. 1998).
The specimen limitations discussed above preclude
our reaching any definitive conclusion on the analysis of
trophoblastic disease samples. However, it appears, as
might be anticipated, that the B152 assay is more sensitive
than the B109 assay in detecting hCG immunoreactivity
in the blood and in the urine of trophoblastic disease
patients, even after treatment.
Acknowledgements
The authors express their appreciation to Dr Laurence
Cole for providing the patient samples and choriocarcinoma hCG (C5) and Dr Amalia Kelly for providing some
clinical specimens at CPMC. We also thank Dr Mark
Sauer for his insight into the clinical applications of this
assay system for programs of assisted reproduction. This
work was made possible by Grant R01 ES07589 from
the National Institute of Environmental Health Sciences,
and the Office of Research on Womens Health, NIH.
Additional support was provided to S B by NIH grants
HD 15454 and AG 13783
References
Berger P, Schwarz S, Spottl G, Wick G & Mann K 1993 Variants of
human chorionic gonadotropin from pregnant women and tumor
patients recognized by monoclonal antibodies. Journal of Clinical
Endocrinology and Metabolism 77 347351.
Birken S, Chen Y, Gawinowicz MA, Lustbader JW, Pollak S, Agosto
G, Buck R & OConnor J 1993 Separation of nicked human
chorionic gonadotropin (hCG), intact hCG, and hCG beta fragment
from standard reference preparations and raw urine samples.
Endocrinology 133 13901397.

G KOVALEVSKAYA

and others

Birken S, Maydelman Y, Gawinowicz MA, Pound A, Liu Y &


Hartree AS 1996 Isolation and characterization of human pituitary
chorionic gonadotropin. Endocrinology 137 14021411.
Bogart MH, Pandian MR & Jones OW 1987 Abnormal maternal
serum chorionic gonadotropin levels in pregnancies with fetal
chromosome abnormalities. Prenatal Diagnosis 7 623630.
Cole LA 1987 The O-linked oligosaccharide structures are strikingly
different on pregnancy and choriocarcinoma hCG. Journal of Clinical
Endocrinology and Metabolism 65 811813.
Cole LA & Kardana A 1992 Discordant results in human chorionic
gonadotropin assays. Clinical Chemistry 38 263270.
Cole LA, Seifer DB, Kardana A & Braunstein GD 1993 Selecting
human chorionic gonadotropin immunoassays: consideration of
cross-reacting molecules in first-trimester pregnancy serum and
urine. American Journal of Obstetrics and Gynecology 168
15801586.
Diaz-Cueto L, Mendez JP, Barrios-de-Tomasi J, Lee JY, Wide L,
Veldhuis JD & Ulloa-Aguirre A 1994 Amplitude regulation of
episodic release, in vitro biological to immunological ratio,
and median charge of human chorionic gonadotropin in
pregnancy. Journal of Clinical Endocrinology and Metabolism 78
890897.
Diaz-Cueto L, Barrios-de-Tomasi J, Timossi C, Mendez JP &
Ulloa-Aguirre A 1996 More in vivo bioactive, shorter-lived human
chorionic gonadotrophin charge isoforms increase at the end of the
first and during the third trimesters of gestation. Molecular Human
Reproduction 2 643650.
Elliott MM, Kardana A, Lustbader JW & Cole LA 1997 Carbohydrate
and peptide structure of the alpha and beta subunits of chorionic
gonadotropin (hCG): characteristics and variants in 32 subunit
preparations from normal and aberrant pregnancy and
choriocarcinoma. Endocrine 7 1532.
Fein HG, Rosen SW & Weintraub BD 1980 Increased glycosylation
of serum human chorionic gonadotropin and subunits from
eutopic and ectopic sources: comparison with placental and
urinary forms. Journal of Clinical Endocrinology and Metabolism 50
11111120.
Haddow JE, Palomaki GE, Knight GJ, Williams J, Miller WA &
Johnson DO 1998 Screening of maternal serum for fetal Downs
syndrome in the first trimester. New England Journal of Medicine 338
955961.
Heinonen S, Ryynanen M, Kirkinen P & Saarikoski S 1996 Elevated
midtrimester maternal serum hCG in chromosomally normal
pregnancies is associated with preeclampsia and velamentous
umbilical cord insertion. American Journal of Perinatology 13
437441.
Hurley TJ, Miller C, OBrien TJ, Blacklaw M & Quirk JG Jr 1996
Maternal serum human chorionic gonadotropin as a marker for the
delivery of low-birth-weight infants in women with unexplained
elevations in maternal serum alpha-fetoprotein. Journal of
MaternalFetal Medicine 5 340344.
Hussa RO 1987 The Clinical Marker hCG. New York: Praeger.
Nemansky M, Thotakura NR, Lyons CD, Reinhold BB, Reinhold
VN & Blithe DL 1998 Developmental changes in the
glycosylation of glycoprotein hormone free alpha subunit
during pregnancy. Journal of Biological Chemistry 15
1206812076.
Nisula BC, Blithe DL, Akar A, Lefort G & Wehmann RE 1989
Metabolic fate of human choriogonadotropin. Journal of Steroid
Biochemistry 33 733737.
OConnor JF, Schlatterer JP, Birken S, Krichevsky A, Armstrong EG,
McMahon D & Canfield RE 1988 Development of highly sensitive
immunoassays to measure human chorionic gonadotropin, its betasubunit, and beta core fragment in the urine: application to
malignancies. Cancer Research 48 13611366.
OConnor JF, Birken S, Lustbader JW, Krichevsky A, Chen Y &
Canfield RE 1994 Recent advances in the chemistry and
Journal of Endocrinology (1999) 161, 99106

105

106

G KOVALEVSKAYA

and others

Early pregnancy hCG

immunochemistry of human chorionic gonadotropin: impact on


clinical measurements. Endocrine Reviews 15 650683.
OConnor JF, Ellish N, Kakuma T, Schlatterer J & Kovalevskaya G
1998 Differential urinary gonadotropin profiles in early pregnancy
and early pregnancy loss. Prenatal Diagnosis 18 12321240.
Odell WD, Griffin J & Sawitzke A 1990 Chorionic gonadotropin
secretion in normal, nonpregnant humans. Trends in Endocrinology
and Metabolism 1 418421.
OLeary P, Nichols C, Feddema P, Lam T & Aitken M 1996 Serum
progesterone and human chorionic gonadotrophin measurements in
the evaluation of ectopic pregnancy. Australian and New Zealand
Journal of Obstetrics and Gynaecology 36 319323.
Onderoglu LS & Kabukcu A 1997 Elevated second trimester human
chorionic gonadotropin level associated with adverse pregnancy
outcome. International Journal of Gynaecology and Obstetrics 56 245249.
Skarulis MC, Wehmann RE, Nisula BC & Blithe DL 1992
Glycosylation changes in human chorionic gonadotropin and free
alpha subunit as gestation progresses. Journal of Clinical Endocrinology
and Metabolism 75 9196.

Journal of Endocrinology (1999) 161, 99106

Ulloa-Aguirre A, Mendez JP, Cravioto A, Grotjan E, DamianMatsumura P & Espinoza R 1990 Studies on the microheterogeneity
of chorionic gonadotrophin secreted by the human cytotrophoblast in
culture. Human Reproduction 5 661669.
Vaitukaitis JL, Braunstein GD & Ross GT 1972 A radioimmunoassay
which specifically measures human chorionic gonadotropin in the
presence of human luteinizing hormone. American Journal of
Obstetrics and Gynecology 113 751758.
Wide L & Hobson B 1987 Some qualitative differences of hCG in
serum from early and late pregnancies and trophoblastic diseases.
Acta Endocrinologica 116 465472.
Wide L, Lee JY & Rasmussen C 1994 A change in the isoforms of
human chorionic gonadotropin occurs around the 13th week of
gestation. Journal of Clinical Endocrinology and Metabolism 78
14191423.

Received 6 July 1998


Revised manuscript received 9 October 1998
Accepted 24 November 1998